Thinking With the Cartesians and Speaking with the Vulgar Extrinsic Denomination in the Philosophy of Antoine Arnauld* Kenneth L. Pearce Trinity College Dublin Abstract Arnauld follows Descartes in denying that sensible qualities like color are modes of external objects. Yet, unlike Malebranche, he resists the apparent implication that ordinary statements like ‘this marble is white’ are false. Arnauld also follows Descartes in saying that we perceive things by having ideas of them. Yet, unlike Malebranche, he denies that this sort of talk implies the existence of intermediaries standing between the mind and its external objects. How can Arnauld avoid these implications? I argue that the answer lies in Arnauld’s sophisticated theory of mental and linguistic representation and, in particular, his account of extrinsic denomination. Keywords: Antoine Arnauld, Extrinsic denomination, ideas, Cartesianism, direct realism, sensible qualities, color In General and Rational Grammar (1660) and Logic or the Art of Thinking (1662), Antoine Arnauld and his collaborators at Port-Royal Abbey put forward a theory of mental and linguistic representation. Their central contention is that the grammatical or syntactic structures of language mirror innate structures of human thought. The structure of thought, in turn, mirrors metaphysical structure in the world. Most crucially, the linguistic relation of predication mirrors the mental act of judgment which in turn mirrors the metaphysical relation of modification. For instance, the sentence ‘the tomato is round’ predicates ‘round’ of ‘the tomato;’ this is mirrored by a mental act of judgment where the idea round is mentally affirmed of the idea the tomato;1 this is in turn mirrored at the metaphysical level where roundness is a mode of an extended substance. There is, however, a problem. Arnauld adheres to a sparse Cartesian ontology on which there are just two kinds of substances, thinking and extended. Furthermore, these substances are properly characterized only by modifications of their principal attributes. That is, every feature of a thinking substance is a way of thinking, and every feature of an extended substance is a way * This is the author’s accepted manuscript. The Version of Record will appear in Journal of the History of Philosophy. 1 In this paper I use quote-names for linguistic items (words, sentences, etc.) and italics for mental items (ideas, judgments, etc.). Where there is no danger of confusion, italics are occasionally also used for emphasis. 1 of being extended. However, Arnauld is committed to the truth of numerous judgments that, given his Cartesian ontology, cannot fit the pattern outlined above. The most widely discussed case is the case of sensible qualities. Nicolas Malebranche was willing to draw the hardline conclusion that, since red is not a mode of extension, judgments like the tomato is red are simply false (Search, 634). Arnauld is unwilling to disturb ordinary language in this way (On True and False Ideas, 173–74). There are also numerous other instances of the same problem. It is often true that the tomato is seen, but being seen is not a mode of extension. Further, it is of crucial importance to the Port-Royal theory of language that “words…indicate what takes place in the mind” (Arnauld and Nicole, Logic, 74). However, words “by their nature” are only “sounds and characters” and therefore belong to the realm of extension (Arnauld and Lancelot, Grammar, 41). The predicate ‘indicate what takes place in the mind,’ which is applied to these extended objects, does not name a mode of extension. A rather different case is of central importance to Arnauld. It is a principle of Cartesian philosophy, and a foundational assumption of the Port-Royal Logic, that “we can have no knowledge of what is outside us except by means of the ideas in us” (25). According to Malebranche and some others, an idea is neither an act of thinking nor an extended object but rather some kind of special ‘representative being.’ Arnauld sees this approach as violating Cartesian metaphysical principles and also undermining our ability to perceive extended objects. He must therefore give an alternative analysis of judgments about ideas. It has usually been thought that he does this by identifying ideas with acts of perceiving. However, as I will argue, this approach does not adequately account for the role of ideas in Arnauld’s theory of language. Arnauld needs to take talk of ideas as objects seriously while avoiding commitment to Malebranchean ‘representative beings.’ It seems, then, that the simplistic view of the mirroring between language, mind, and world outlined above applies only to a quite restricted class of cases. In this paper, I show that Arnauld’s philosophy contains a sophisticated, unified, and powerful response to this family of problems that has hitherto gone unrecognized: the theory of extrinsic denomination. Extrinsic denomination is the naming of an object with “names derived from the actions of something else” (Arnauld and Nicole, Logic, 32). That is, in extrinsic denomination, a name is truly predicated of one thing in virtue of a mode possessed by some other thing. The most prominent examples are what Roderick Chisholm2 called ‘converse intentional properties.’ For instance, in a well-known example derived from Descartes, when we judge that the sun is perceived, we affirm being perceived of the sun. But our idea of being perceived signifies the mode perceiving, which modifies the perceiver and not the sun. I begin, in section 1, by giving a more detailed exposition of how the Port-Royal theory deals with straightforward cases like ‘the tomato is round.’ Then, in section 2, I introduce the Port-Royal theory of extrinsic denomination. In the remaining three sections, I argue that Arnauld takes numerous philosophically crucial terms, including ‘idea,’ ‘sign,’ ‘word,’ and ‘red’ 2 “Converse Intentional Properties.” 2 to denominate objects only extrinsically, and show how this account provides a response to the difficulties outlined above. 1. The Port-Royal Theory of Language The most basic assumption of the Port-Royal theory of language is that “words are distinct and articulated sounds that people have made into signs to indicate what takes place in the mind” (Arnauld and Nicole, Logic, 74).3 ‘What takes place in the mind’ includes four operations, treated in the four parts of the Logic: conceiving, judging, reasoning, and ordering. Our present concern includes the first two. To conceive is simply to have a thought about an object. Such an object of thought is called an ‘idea’ and is typically signified in language by a noun (74–5). Judging is the operation whereby these ideas are assembled into truth-evaluable propositions. In the Logic, this process is explained as follows: After conceiving things by our ideas, we compare these ideas and, finding that some belong together and others do not, we unite or separate them. This is called affirming or denying, and in general judging. This judgment is also called a proposition, and it is easy to see that it must have two terms. One term, of which one affirms or denies something, is called the subject; the other term, which is affirmed or denied, is called the attribute or Praedicatum. It is not enough to conceive these two terms, but the mind must connect or separate them…this action of the mind is indicated in discourse by the verb “is,” either by itself when we make an affirmation, or with a negative particle in a denial. (82) According to the Port-Royal view the judgment always has three components: a subject idea, a predicate idea, and the mental act of judging which unites them in a truth-evaluable proposition. Although these are not always expressed in speech, this is merely a matter of abbreviation (79– 80). The Grammar and Logic both deal at length with complex sentences and propositions, but these are always taken to involve the affirmation of some (often quite complex) predicate idea of some (again, often quite complex) subject idea. Although this is occasionally obscured by the surface structure, subject-predicate structure is universal. The Port-Royalists adopt the traditional view that although “there is a kind of truth in things with respect to God’s mind, whether people think of it or not…there can be falsity only relative to the human mind or to some other mind subject to error, that falsely judges that a thing is what it is not” (Arnauld and Nicole, Logic, 91–92).4 In other words, truth and falsity 3 Compare Nicole, Arnauld, and Renaudot, Perpétuité, 2:81: “It is certain that if people could immediately see what passes in the mind and in the heart of others, they would not speak at all, and words would have no use, because they have no [use] other than to make known our thoughts to those who we suppose do not know them.” (Translations from the Perpetuity are my own.) 4 Compare Aquinas, On Truth, question I, article x. 3 necessarily involve a cognitive agent getting things right or wrong, and this is why it takes an agent’s act of judging to form a truth-evaluable proposition.5 This view implies that judging somehow involves taking the world to be a certain way. In introducing the act of judging, Arnauld and Nicole said that it involved “compar[ing]…ideas and, finding that some belong together and others do not…unit[ing] or separat[ing] them” (82). But what is it for ideas to ‘belong together?’ The closest the Logic comes to a definition of truth is the remark that “Since every proposition indicates a judgment we make about things, it is true when this judgment conforms to the truth and false when it does not” (84).